Internet DRAFT - draft-mcfadden-smart-rfc3552-textual-research
draft-mcfadden-smart-rfc3552-textual-research
SMART M. McFadden
Internet-Draft internet policy advisors ltd
A. Mills
UWE - Bristol
Intended status: Informational September 9, 2020
Expires: March 9, 2021
Textual Analysis Methodology for Security Considerations Sections
draft-mcfadden-smart-rfc3552-textual-research-02.txt
Status of this Memo
This Internet-Draft is submitted in full conformance with the
provisions of BCP 78 and BCP 79.
Internet-Drafts are working documents of the Internet Engineering
Task Force (IETF), its areas, and its working groups. Note that
other groups may also distribute working documents as Internet-
Drafts.
Internet-Drafts are draft documents valid for a maximum of six
months and may be updated, replaced, or obsoleted by other documents
at any time. It is inappropriate to use Internet-Drafts as
reference material or to cite them other than as "work in progress."
The list of current Internet-Drafts can be accessed at
http://www.ietf.org/ietf/1id-abstracts.txt
The list of Internet-Draft Shadow Directories can be accessed at
http://www.ietf.org/shadow.html
This Internet-Draft will expire on March 9, 2021.
Copyright Notice
Copyright (c) 2020 IETF Trust and the persons identified as the
document authors. All rights reserved.
This document is subject to BCP 78 and the IETF Trust's Legal
Provisions Relating to IETF Documents
(http://trustee.ietf.org/license-info) in effect on the date of
publication of this document. Please review these documents
carefully, as they describe your rights and restrictions with
respect to this document. Code Components extracted from this
document must include Simplified BSD License text as described in
McFadden, Mills Expires September 9, 2020 [Page 1]
Internet-Draft RFC3552 Research Methodology September 2020
Section 4.e of the Trust Legal Provisions and are provided without
warranty as described in the Simplified BSD License.
Abstract
[RFC3552] provides guidance to authors in crafting RFC text on
Security Considerations. The RFC is more than fifteen years old.
With the threat landscape and security ecosystem significantly
changed since the RFC was published, RFC3552 is a candidate for
update. This draft proposes that, prior to drafting an update to
RFC3552, an examination of recent, published Security Considerations
sections be carried out as a baseline for how to improve RFC3552. It
suggests a methodology for examining Security Considerations
sections in published RFCs and the extraction of both quantitative
and qualitative information that could inform a revision of the
older guidance. It also reports on a recent experiment on textual
analysis of sixteen years of RFC Security Consideration sections.
Table of Contents
1. Introduction...................................................3
2. Conventions used in this document..............................3
3. Motivation.....................................................4
3.1. Non-goals and scoping.....................................5
3.2. Research Group............................................5
4. Goals for Surveying Existing Security Considerations Sections..5
5. Methodology....................................................5
5.1. Methodology Overview......................................5
5.2. Quantitative Methodology..................................6
5.3. Qualitative Methodology...................................7
5.4. Implications of the Size of n-set.........................7
6. Experimental Activity..........................................8
6.1. Experiment Methodology....................................8
6.2. Stopword List.............................................8
6.3. Resulting Characterization...............................10
6.4. Indicative Results.......................................11
6.4.1. Top Ten Word Counts in Four Sample Years............11
6.4.2. Top Ten Word Counts Without RFC2119 Words in Four
Sample Years...............................................12
6.4.3. Normative RFC2119 Words in Security Considerations..12
7. Security Considerations.......................................13
8. IANA Considerations...........................................13
9. References....................................................13
9.1. Normative References.....................................13
9.2. Informative References...................................13
Appendix A. Document History.....................................14
McFadden, Mills Expires March 9, 2021 [Page 2]
Internet-Draft RFC3552 Research Methodology September 2020
Appendix B. 75 Most Common Words in Security Considerations Sections
.................................................................15
1. Introduction
[RFC2223] requires that all RFCs have a Security Consideration
section. The motivation of the section is both to encourage RFC
authors to consider security in protocol design and to inform
readers of relevant security issues. RFC3552 was published in July
of 2003 to give guidance to RFC authors on how to write a good
Security Considerations section. It is structured in three parts: a
tutorial and definitional section, then a series of guidelines, and
finally a series of examples.
It is possible to observe that the Internet security landscape has
changed significantly since the publication of RFC3552. Rather than
an immediate attempt to draft and discuss a revision to the older
RFC, it may be prudent to learn from the experience of more than
fifteen years of documents published since RFC3552 was approved for
publication.
It is possible that an examination of published Security
Considerations sections of existing documents could give both
quantitative and qualitative insight on how to proceed with a newer
version of the Security Considerations guidelines. The motivation is
to inform any discussion of a revision with quantitative and
qualitative data gleaned from years of published RFCs.
This document proposes a methodology for such research.
This scope of this proposal is for the research itself. Discussion
of relevant issues, document organization and revised content for a
revision of RFC3552 is out of scope. Instead, the motivation is to
guide a piece of research that would later form part of the
foundation for a discussion of a revision to RFC3552.
2. Conventions used in this document
The key words "MUST", "MUST NOT", "REQUIRED", "SHALL", "SHALL NOT",
"SHOULD", "SHOULD NOT", "RECOMMENDED", "MAY", and "OPTIONAL" in this
document are to be interpreted as described in RFC 2119 [RFC2119].
In this document, these words will appear with that interpretation
only when in ALL CAPS. Lower case uses of these words are not to be
interpreted as carrying significance described in RFC 2119.
McFadden, Mills Expires March 9, 2021 [Page 3]
Internet-Draft RFC3552 Research Methodology September 2020
3. Motivation
Since 1998, all RFCs have been required to have a Security
Considerations section. The authors of RFC3552 observed that
"historically, such sections have been relatively weak." The
motivation for RFC3552 was, in part, to improve the quality of
Security Considerations sections.
Today the Internet threat model, the landscape of attacks, and our
understanding of how to craft protocols that are more robust and
resilient has changed significantly. Experience in both protocol
design and implementation has greatly improved our understanding of
the security implications of choices made during protocol design.
It is possible that a revision of RFC3552, reflecting the changes to
the Internet and our understanding of the evolved security landscape
and threat model, is appropriate. The IAB is currently examining and
reassessing the Internet's threat model [1].
The IAB has previously discussed a potential revision to RFC3552 in
its report from the Strengthening the Internet (STRINT) Workshop. In
section 2 of [RFC7687], the editors report that "...the IETF may be
in a position to start to develop an update to BCP 72 [RFC3552],
most likely as a new RFC enhancing that BCP and dealing with
recommendations on how to mitigate PM and how to reflect that in
IETF work."
If a revision were to be contemplated, it would be useful to learn
from the body of experience of crafting Security Considerations
sections in recent years. That body of experience could inform the
discussion of what makes up a good Security Considerations section
by collecting real-world data from existing RFCs. It would be
possible to have a survey of the existing Security Considerations
sections in published RFCs. The data collected from that survey
could provide one source of information for discussion of how to
improve upon RFC3552 in the current environment.
For such a survey to be successful, an outline of some basic goals
and a methodology would be required. This document provides those
goals and methodology. The intent is that individuals or
organizations could then carry out such a survey, publish the
results and use that data to inform any discussion of a potential
3552bis.
This draft also documents the results of a recent experiment to
conduct an automated survey of words in Security Considerations
sections.
McFadden, Mills Expires March 9, 2021 [Page 4]
Internet-Draft RFC3552 Research Methodology September 2020
3.1. Non-goals and scoping
This document specifically does not make suggestions for changes to
RFC3552. It also does not identify changes to the Internet threat
model or the general security landscape that has changed since that
RFC has been published.
The scope of this document is to provide a basic set of goals for
research on existing Security Considerations sections and establish
a methodology for conducting that research.
3.2. Research Group
The research work suggested in this document was envisioned and
intended to be carried out as a research activity of the proposed
Stopping Malware and Researching Threats (SMART) research group in
the IRTF. The work could also be conducted independently and
submitted as an Independent Submission in the IETF.
4. Goals for Surveying Existing Security Considerations Sections
A cursory examination of recent years' Security Considerations
sections shows that authors publish a wide variety of these
sections. This is natural since the RFC series has a diverse set of
purposes and readership.
However, even a cursory examination shows that published Security
Considerations sections have some clear characteristics. Identifying
useful characteristics and then surveying the existing base of
published RFCs may provide a useful base of information for a later
discussion of revising RFC3552.
The goal of surveying existing Security Considerations sections is
to provide quantitative and qualitative data, from existing,
published RFCs, that can be used to inform a discussion of revising
RFC3552.
5. Methodology
5.1. Methodology Overview
The survey of existing Security Considerations sections would
examine a subset of RFCs published since the publication of RFC3552.
RFCs obsoleted by later publications, RFCs that are reports from IAB
activities and IETF, IRTF, and IESG administrative RFC are omitted
from consideration.
McFadden, Mills Expires March 9, 2021 [Page 5]
Internet-Draft RFC3552 Research Methodology September 2020
The survey should select a specific timeframe, across which, all
RFCs published in that period are examined.
The examination proceeds in two parts: a quantitative examination of
the Security Considerations sections and then a qualitative
examination.
As an example, the quantitative examination might survey and collect
data on the source of the RFC (e.g. Security Area, Routing Area,
Transport Area), whether the RFC extends the Security Considerations
section of a previously published document, the wordcount of the
section, and the existence of specific keywords.
The qualitative analysis might group Security Considerations
sections by particular characteristics - those characteristics being
discovered, in part, during an initial examination of the published
documents.
5.2. Quantitative Methodology
Once the set of RFCs (where the size of the set is said to be n-set)
to be considered is established, the quantitative analysis proceeds
as follows for each item in the set:
o recording the date of publication
o recording the source of the original draft
o recording the category of the RFC (e.g. Informational, etc.)
o recording the size of the Security Considerations section in
words and paragraphs
o recording whether or not the section updates or extends the
Security Considerations section of a previously published
document
o record whether or not examples exist in the Security
Considerations section
o record whether or not example code appears in the Security
Considerations section
o extracting the text and creating a new text removing the 100 most
common English words
McFadden, Mills Expires March 9, 2021 [Page 6]
Internet-Draft RFC3552 Research Methodology September 2020
o against the new text created in the step above, perform text
analytics - for instance, create a count of the number of
occurrences of expected keywords
The result would be a series of metrics for n-set that establish
certain characteristics of the Security Considerations sections of
published RFCs. Once the quantitative data was gathered, further
analysis of the data could be conducted (for instance, finding
relationships between certain features of the RFCs).
5.3. Qualitative Methodology
The documents could also be assigned qualitative characteristics as
a result of the survey. For instance, based on characteristics of
the document, the Security Considerations could be characterized as
"extensive" or "limited."
It is also clear that analysis of the Security Considerations could
lead to other groupings. For instance, an analysis of recent RFCs
shows that those documents which focus on cipher suites have quite
different security considerations sections compared to those that
extend and existing protocol. Identification of those
characteristics might be possible during an initial survey. In
another case, those characteristics might emerge during the survey
execution.
5.4. Implications of the Size of n-set
Since part of the execution of the survey has to be done via human
intervention, the size of n-set has an effect on whether or not
volunteers or organizations take on the effort. While it would be
helpful to have as large a sample size as possible for the
collection of data to support the analysis. It may be necessary to
limit the size of n-set in practice.
One way to do this is to limit the range of dates for the RFCs being
analyzed. A cursory, initial examination of Security Considerations
sections seems to indicate that, in recent years, a clear set of
prototypical security considerations sections has emerged and that
there are distinct type of sections. By limiting the RFCs for the
set of considered document to a specific, recent timeframe the goal
is to focus the analysis on recent practice in crafting Security
Considerations sections and moving them through the document
approval process.
Another approach to solving the potential problem of the size of n-
set is to incorporate a sampling regime for the selection of RFCs to
McFadden, Mills Expires March 9, 2021 [Page 7]
Internet-Draft RFC3552 Research Methodology September 2020
be examined. This would be a meaningful approach in the event where
the timeframe was extended, but where it was still desirable to
reduce the size of n-set.
This proposal suggests to use the timeframe limitation but not
incorporate sampling.
6. Experimental Activity
One of the authors has conducted an experiment that is consistent
with many of the features of the methodology in Section 5 above.
This experiment uses a pair of Python scripts to extract the
Security Considerations sections from historic RFCs and then parse
those sections to get word frequency information from those Security
Considerations.
The initial experiment was motivated by a desire to see if one could
detect changes in Security Considerations section wording after
significant security incidents in the public Internet. In
particular, the experiment was designed to detect changes in the
frequency of words over time.
6.1. Experiment Methodology
The RFC series was grouped into input files based on the year of
publication of the RFC.
Using HTML versions of the RFC series document as input, these were
put through an open source parser. The parser then identified the
words "Security Consideration" or "Security" in header text. It then
output that text to a temporary file in UTF-8 encoding until the
parser encountered the next section.
The parser removed non-textual material from the temporary files
including hyphens, RFC references, anchor URLs, other sections
references, standalone letters and other characters that were not
words.
It then built a frequency list for all words not in a designated
list of words not to be counted. This list is a variable and could
be changed to include, or exclude, words from the designated list.
6.2. Stopword List
The following list of words were used as the designated list of
words not to be counted:
McFadden, Mills Expires March 9, 2021 [Page 8]
Internet-Draft RFC3552 Research Methodology September 2020
. Also
. Could
. Would
. However
. One
. See
. Use
. Therefore
. Discussed
. New
. March
. Type
. Even
. Following
. Without
. Bradner
. Using
. Described
. Might
. Thus
. Two
. Since
. Different
. Number
McFadden, Mills Expires March 9, 2021 [Page 9]
Internet-Draft RFC3552 Research Methodology September 2020
. Via
. Mechanism
. Used
. Tl
. Header
. Field
. Name
. Sent
6.3. Resulting Characterization
The result of this experiment is a pair of files for each year
starting in 2003. The two files for each year are:
. A word frequency file sorted by the number of times a
particular word appears in the Security Considerations section
of RFCs published in that year; and,
. An RFC Count file that counts how many times each RFC was
mentioned within the Security Considerations sections.
The idea behind the second file was to see if there was a trend or
change in the RFCs cited and what this might suggest or say in
regards to the content of these sections. For example in 2004 the
highest referenced RFC was [RFC3410] Applicability Statements for
SNMP in 2009 it was [RFC4301] Security Architecture for IP though
[RFC3410] was also referenced a high number of times.
As [RFC4301] came out in 2005 we would not expect it to be
referenced in 2004, but the reference count in 2009 could indicate
that there were a number of RFCs which likely simply referred to the
Security Considerations Section of this RFC in a line similar to
"this extends the security consideration of <insert RFC here>." This
could then be used to help narrow down qualitative focus on this
highly referenced RFCs and to also see if in some cases lip service
is all that is occurring within other Security Considerations
Sections.
Another result, included with the word frequency file, is a list of
words similar to the word "security" based on context analysis. This
McFadden, Mills Expires March 9, 2021 [Page 10]
Internet-Draft RFC3552 Research Methodology September 2020
is another indicator that can be used to look at how the language of
the RFC series is changing. For example looking at 2004 the most
similar words are:
. Used, ipsec, mode, authentication, implementation, message,
may, watcher, method and block.
In 2009:
. Message, attacker, syslog, used, attack, information,
transport, gruu, may and case.
Yet another result was a file that provides comparative data for
word counts in the Security Considerations and Privacy
Considerations sections of published RFCs. The result provides a
look at whether the length of those sections might have changed over
time.
A final result was a Frequency count over the entire period examined
for Internet Standards, BCPs, and Proposed Standards. This result
gives indication of whether or not the average length of these
sections has changed - either over time, or in response to specific
security incidents on the public Internet.
6.4. Indicative Results
This draft is focused on proposing a methodology and not on the
experiment being reported on here. However, there are some
indicative results that may be of use as a future methodology is
considered. It is worth observing that the original motivation for
the experiment - to see if Security Considerations sections changed
in the face of security-related events on the public Internet -
showed that no significant re-wording took place over the timeframe
studied.
6.4.1. Top Ten Word Counts in Four Sample Years
Choosing four sample years - 2019 2014 2009 and 2004 as examples,
the experiment found the following most frequent words in Security
Considerations sections (the lists are in most frequent to tenth
most frequent).
. 2019 - security, server, data, message, may, network, attack,
information, client, xmpp-grid
. 2014 - security, information, attack, message, may, used,
server, data, authentication, network
McFadden, Mills Expires March 9, 2021 [Page 11]
Internet-Draft RFC3552 Research Methodology September 2020
. 2009 - security, may, message, address, attack, used, packet,
protocol, network, information
. 2004 - security, may, key, authentication, object, used,
information, message, attack, access
6.4.2. Top Ten Word Counts Without RFC2119 Words in Four Sample Years
Taking the same data and removing the normative words that are
defined in RFC2119 leads to slightly different results.
. 2019 - security, server, data, message, network, attack,
information, client, xmpp-grid, document
. 2014 - security, information, message, used, server, data,
authentication, network, attacker
. 2009 - security, message, address, attack, used, packet,
protocol, network, information, object
. 2004 - security, key, authentication, object, used,
information, message, attack, access, user
6.4.3. Normative RFC2119 Words in Security Considerations
The word MAY always appears more often than any other RFC2119 word
in Security Considerations sections. The word MUST most often
appears after MAY and is often in the top 15 words sorted by
frequency.
However, the word SHOULD hardly ever appears in the top 100 most
frequent words for any year of published RFCs.
Most Frequent Words in Proposed Standards Security Considerations
Over the entire period 2003-2019, the most frequent non-normative
words in Security Considerations sections was:
. Security, message, attack, server, information, key,
authentication, network, protocol, client
A list of the 75 most commonly, non-normative words is provided in
Appendix B.
McFadden, Mills Expires March 9, 2021 [Page 12]
Internet-Draft RFC3552 Research Methodology September 2020
7. Security Considerations
This document describes goals and a methodology for surveying the
existing body of Security Considerations in published RFCs. It does
not create, extend or modify any protocols. Its intent is to provide
a foundation for a data-driven discussion of the guidelines for
writing a Security Considerations section in an RFC.
8. IANA Considerations
Upon publication, this document has no required actions for IANA.
9. References
9.1. Normative References
[RFC2119] Bradner, S., "Key words for use in RFCs to Indicate
Requirement Levels", BCP 14, RFC 2119, March 1997.
[RFC2223] Postel J. and Reynolds J., ISI, "Instructions to RFC
Authors", RFC2223, October 1997.
[RFC3552] Rescorla E. and Korver B.(Editors), "Guidelines for
Writing RFC Text on Security Considerations", BCP 72,
RFC3552, July 2003.
[RFC7687] Farrell S., Wenning R., Bos B., Blanchet M. and Tschofenig
H., "Report from the Strengthening the Internet (STRINT)
Workshop, RFC 7687, December 2015
9.2. Informative References
[1] Model-t -- Discussions of changes in Internet deployment
patterns and their impact on the Internet threat model,
https://www.ietf.org/mailman/listinfo/model-t
[2] Acknowledgments
This document was prepared using 2-Word-v2.0.template.dot.
McFadden, Mills Expires March 9, 2021 [Page 13]
Internet-Draft RFC3552 Research Methodology September 2020
Appendix A. Document History
-00
Initial Internet Draft
-01
Section 6 and Appendix B are added. Significant editing of Section 3
on Motivation and Section 5 on Methodology. Several typos fixed.
-02
This draft was presented at the combined RIPE MAT and IRTF MAPRG
meeting held on August 8,2020. Many of the comments during that
session focused on the fact that words like privacy and surveillance
did not appear in the lists of most commonly used words. After
presentation, a quick check indicates that those words, after the
publication of [RFC7258] do not appear int eh Security
Considerations sections of those documents.
The MAPRG presentation of this draft is available at the following
URL:
https://trac.ietf.org/trac/irtf/wiki/map
McFadden, Mills Expires March 9, 2021 [Page 14]
Internet-Draft RFC3552 Research Methodology September 2020
Appendix B. 75 Most Common Words in Security Considerations Sections
Over the entire period 2003-2019, the 75 most frequent words in
Security Considerations sections was (in order by frequency):
security, message, attack, data, used, may, authentication, key,
access, protocol, information, must, address, transport, process,
model, client, server, network, ipfix, tl, user, traffic, packet,
object, operation, control, service, ipp, example, document,
implementation, measurement, collecting, secure, header, attacker,
identity, value, job, need, support, snmp, provide, printer, uri,
certificate, authenticated, possible, name, content, source,
connection, field, set, system, dtls, cause, sensitive, domain,
provides, configuration, router, privacy, protection, peer, nacm,
layer, ip, device, exporting, within, request, large, and signature.
McFadden, Mills Expires March 9, 2021 [Page 15]
Internet-Draft RFC3552 Research Methodology September 2020
Authors' Addresses
Mark McFadden
Internet policy advisors ltd
Madison Wisconsin US
Email: mark@internetpolicyadvisors.com
Alan Mills
University of the West of England, Bristol
Bristol BS16 1QY United Kingdom
Email: Alan2.Mills@live.uwe.ac.uk
McFadden, Mills Expires March 9, 2021 [Page 16]
